Ticket #15091: wordpress-importer.diff

Index: wordpress-importer.php
===================================================================
--- wordpress-importer.php      (revision 298755)
+++ wordpress-importer.php      (working copy)
@@ -463,15 +463,8 @@
                $post_content = str_replace('<br>', '<br />', $post_content);
                $post_content = str_replace('<hr>', '<hr />', $post_content);
 
-               preg_match_all('|<category domain="tag">(.*?)</category>|is', $post, $tags);
-               $tags = $tags[1];
+               preg_match_all('|<category\s*(?:\bdomain="(.*?)"\s*)?(?:\bnicename="(.*?)")?>(?:<!\[CDATA\[)?(.*?)(?:\]\]>)?</category>|', $post, $terms, PREG_SET_ORDER);
 
-               $tag_index = 0;
-               foreach ($tags as $tag) {
-                       $tags[$tag_index] = $wpdb->escape( html_entity_decode( str_replace(array( '<![CDATA[', ']]>' ), '', $tag ) ) );
-                       $tag_index++;
-               }
-
                preg_match_all('|<category>(.*?)</category>|is', $post, $categories);
                $categories = $categories[1];
 
@@ -556,26 +549,30 @@
                        }
 
                        // Add terms.
-                       if (count($tags) > 0) {
-                               $post_tags = array();
-                               foreach ($tags as $tag) {
-                                       if ( '' == $tag )
+                       if (count($terms) > 0) {
+                               $post_terms = array();
+                               foreach ($terms as $term) {
+                                       if ( '' == $term[1] || 'category' == $term[1])
                                                continue;
-                                       $slug = sanitize_term_field('slug', $tag, 0, 'post_tag', 'db');
-                                       $tag_obj = get_term_by('slug', $slug, 'post_tag');
-                                       $tag_id = 0;
-                                       if ( ! empty($tag_obj) )
-                                               $tag_id = $tag_obj->term_id;
-                                       if ( $tag_id == 0 ) {
-                                               $tag = $wpdb->escape($tag);
-                                               $tag_id = wp_insert_term($tag, 'post_tag');
-                                               if ( is_wp_error($tag_id) )
+                                       $slug = $term[2];
+                                       echo $slug;
+                                       $domain = $term[1];
+                                       if ($domain == 'tag') {
+                                               $domain = 'post_tag';
+                                       }
+                                       $term = $wpdb->escape($term[3]);
+                                       $term_id = term_exists( $term, $domain );
+                                       if ( $term_id || $term_id != 0) {
+                                               echo $term_id;
+                                       } else {
+                                               $term_id = wp_insert_term($term, $domain);
+                                               echo $term_id;
+                                               if ( is_wp_error($term_id) )
                                                        continue;
-                                               $tag_id = $tag_id['term_id'];
+                                               $term_id = $term_id['term_id'];
                                        }
-                                       $post_tags[] = intval($tag_id);
+                                       wp_set_post_terms($post_id, $term, $domain,true);
                                }
-                               wp_set_post_tags($post_id, $post_tags);
                        }
                }